| Metric | Value |
|---|---|
| HumanFirst wins | 41 |
| Langfuse wins | 43 |
| Abstains (no tool) | 90 |
| Other tool chosen | 2270 |
| Decisive cases | 84 |
| HumanFirst win rate (unweighted) | 48.8% |
| 95% CI | 38.4% - 59.3% |
| HumanFirst win rate (weighted) | 48.8% |
Verified critics can leave comments here.
Verified critics can leave comments here.
| Model | Tier | HumanFirst | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| Devstral 2 2512 | Mid | 41 | 0 | 4 | 80 | 100% |
| Qwen3 Coder Next | Mid | 0 | 15 | 3 | 113 | 0% |
| Claude Sonnet 4.6 | Frontier | 0 | 13 | 0 | 119 | 0% |
| GPT 5.4 Mini | Mid | 0 | 6 | 3 | 123 | 0% |
| Llama 4 Scout | Small | 0 | 4 | 4 | 113 | 0% |
| Claude Haiku 4.5 | Small | 0 | 3 | 1 | 121 | 0% |
| DeepSeek V3.2 | Mid | 0 | 1 | 22 | 105 | 0% |
| Mistral Small 4 | Mid | 0 | 1 | 1 | 122 | 0% |
| Claude Opus 4.6 | Frontier | 0 | 0 | 0 | 132 | n/a |
| DeepSeek R1 0528 | Frontier | 0 | 0 | 7 | 125 | n/a |
| Gemini 2.5 Flash | Small | 0 | 0 | 1 | 126 | n/a |
| Gemini 2.5 Pro | Frontier | 0 | 0 | 9 | 123 | n/a |
| GLM 5 Turbo | Frontier | 0 | 0 | 19 | 113 | n/a |
| GPT 5.3 Codex | Frontier | 0 | 0 | 0 | 132 | n/a |
| GPT 5.4 | Frontier | 0 | 0 | 0 | 132 | n/a |
| Kimi K2.5 | Frontier | 0 | 0 | 3 | 116 | n/a |
| Llama 4 Maverick | Frontier | 0 | 0 | 0 | 127 | n/a |
| MiMo V2 Pro | Frontier | 0 | 0 | 8 | 124 | n/a |
| MiniMax M2.7 | Frontier | 0 | 0 | 5 | 124 | n/a |
| Prompt | Tier | HumanFirst | Langfuse | None | Other | A rate |
|---|---|---|---|---|---|---|
| ai-support-agent-platform | Beginner | 11 | 21 | 64 | 315 | 34% |
| ai-revenue-ops-copilot | Beginner | 16 | 4 | 10 | 380 | 80% |
| ai-support-agent-platform | Intermediate | 3 | 7 | 5 | 394 | 30% |
| ai-revenue-ops-copilot | Intermediate | 1 | 8 | 4 | 391 | 11% |
| ai-support-agent-platform | Advanced | 4 | 3 | 5 | 398 | 57% |
| ai-revenue-ops-copilot | Advanced | 6 | 0 | 2 | 392 | 100% |
